The prostheses currently available on the market, e.g. arm, leg, hand, and foot prostheses, all lack conscious sensory feedback.
Congenital or traumatic amputation of e.g. a hand is a catastrophe with impact not only on occupational life and leisure activities but also on identity, social interaction and the quality of life. To compensate for the cosmetic defect various types of cosmetic hand prostheses are available, e.g. to provide a functionally useful substitution for the hand myoelectric prostheses are used, although with considerable restrictions for many patients.
Myoelectric prostheses are motorised devices which are capable of opening or closing the hand on the basis of mass EMG-signals (electromyografic signals) monitored by one surface electrode arranged on top of the extensor muscles of the amputation stump and a surface electrode arranged on top of the flexor muscles. Myoelectric hand prostheses are available on the market. Although such neurally controlled myoelectric prostheses are used by some patients there are several disadvantages in such prostheses e.g. lack of capacity for delicate movements, the prostheses usually allowing only two degrees of freedom, such as opening and closure of the hand, and lack of conscious sensory feedback. Additionally, factors like weight and limited battery lifetime are also disadvantageous in such prostheses.
Ongoing developmental research work, including the combined use of multiple EMG recording sites and artificial neural networks may soon form a base for neural control of more advanced hand prostheses, providing a large number of degrees of freedom as described by Sebelius et al. (2005). However, the lack of conscious sensory feedback still represents a fundamental major problem. The current innovation specifically addresses this problem.
The function of e.g. the human hand is based upon a complex interplay between efferent signals for motor control and afferent signals from various sensory system providing sensory feedback (Scott, 2004). The central nervous system plays a crucial role for the processing and integration of sensory input and motor output. Although the spinal cord and brain stem support “automatic functions” like limb reflexes, the motor cortex of the brain supports and controls voluntary and planned motor tasks such as reaching for an object or performing learning motor behaviour (Moran and Schwartz, 1999; Gribble and Scott, 2002). Although the planning of motor activities involves several cortical regions such as the frontal parietal region—premotor cortex (Rizzolatti and Luppino, 2001; Battaglia-Mayer et al., 2003) the execution of goal-directed and skill motor tasks is based on neural activity in the primary motor cortex Ml (Porter and Lemon, 1993). The activity of Ml is to a great extent modulated and influenced by sensory feedback from several senses such as touch, vision, hearing and proprioception. Many nerve cells in Ml respond strongly to cutaneous stimulation of the hand (Picard and Smith, 1992), a fact which strongly reflects how grip functions of the hand are dependent on skin contact and cutaneous stimulation constituting a sensory feedback (Scott, 2004).
The sense of touch is essential for making a hand “belonging to the body”. A hand without sensory functions is perceived as a foreign body and may even be denied by the owner (Ramachandran and Blakeslee, 1998). In addition, regulation of grip force and execution of delicate motor tasks in the hand are dependent on a sensory input from the hand to the central nervous system. Typically the sensibility of the glabrous skin of the hand is based on four types of mechanoreceptors, localised in subepidermal and subcutaneous areas, and responding to static pressure or vibrotactile stimuli (Johansson and Vallbo, 1979; Johansson and Birznieks, 2004). Among receptors responding to vibration are Meissner's corpuscles, located in the subepidermal papillae, with small receptive fields (Fast Adapting—type I—FA I receptors) and Pacini's corpuscles, located in subcutaneous layers possessing large receptive fields (Fast Adapting type II—FA II receptors-). The Merkel cells, located just beneath the epithelium, respond to static measure and have small receptive fields (Slowly Adapting—type I—SA I receptors). Ruffini's organ, located subcutaneously, responds mainly to stretching (SA II receptor). The extremely well developed tactile functions of the hand make the hand a sensing organ. The sense of touch is essential for exploring the surrounding world. In active touch we can, without using vision, recognise and identify textures, shapes and forms of small items (Klatsky et al., 1987; Katz, 1989). The well developed sensory function of the hand is reflected in the representational area of the hand in somatosensory cortex, occupying a very substantial part (Merzenich and Jenkins, 1993; Kaas, 1997).
Various principles for providing sensory feedback in hand prostheses have been tried over the years (see reviews by (Riso, 1999; Lundborg and Rosen, 2001). Three different approaches have been used:                1) sensory substitution (use of an intact sensory system to replace the missing one);        2) direct stimulation of intact nerves;        3) transferred cutaneous sensation implying stimulation of intact skin in a remote area of the body.        
Sensory substitution, using an intact sensory system, is automatically used by amputees using myoelectric prostheses, which utilise vision to guide the movements of the prosthetic hand. The use of hearing as substitution for missing sensation has been described as an effective strategy in major nerve injuries leaving the hand devoid of sensation and has also been tried in hand prostheses as described in WO9848740 (A1) by Lundborg. According to this principle electrical signals, elicited by vibrotactile stimuli and recorded by miniature microphones at fingertip level when fingers are touched, are fed to a stereo processor which transfers weighted signals to earphones applied to the left and right ears, making possible spatial resolution of the hand. In this way the “friction sound” associated with active touch of various textures can be recognised and easily associated with touch of specific textures (Lundborg et al., 1999).
It has been demonstrated that sensations can be evoked when sensory nerves of the residual limb are electrically stimulated (Anani and Korner, 1979). Some topographic discrimination can be achieved by selective stimulation of different fascicles using percutaneous stimulatory electrodes (Anani and Korner, 1979). However, the spatial resolution for fibre activation is low using this technique and the quality of perceptived sensation is low.
To improve the principle of direct nerve stimulation various techniques for micro stimulation of tactile afferents have been proposed. More natural sensations can be evoked when individual cutaneous afferent fibres are electrically activated in isolation (Riso, 1999). To further refine the contact nerve tissues, stimulatory devices of various types of nerve interfaces have been used in laboratory environment (Riso, 1999). Some of these techniques require growth of regenerating axons into a stimulation device. For instance, using “sieve electrodes”—regenerating axons from a transected nerve is directed to penetrate the matrix of pores to make appropriate electrode contacts (Wallman et al., 2001; Ceballos et al., 2002). With this principle groups of axons can penetrate separate perforations in the chip and the electrode contact can address either individual pores or separate sections of the transversely positioned sieve electrode. Although the foreign body reaction induced by silicone sieve electrodes is discrete and the long term effects of such electrode implantation are not known. Selective fascicular stimulation can also be achieved by the use of multipolar cuff electrodes (Rodriguez et al., 2000; Navarro et. al., 2001). Another concept for a type of nerve regeneration interface is to introduce, coaxially, platinum-iridium wires or analogous devices into the lumen of a regeneration tube (Edell, 1986; Riso, 1999). Another technique is to use a “brush array” of micromachined needle electrodes inserted into a peripheral nerve (Rutten et al., 1999). Peripheral nerves can also be interfaced by introducing longitudinal fine wires into the nerve trunk (Dhillon et al., 2004) or into separate nerve fascicles (Yoshida and Horch, 1993). The theoretical advantage of the nerve-interface strategies discussed above is that sensory stimuli can be directly transferred into peripheral nerves and can thereby reach CNS. However, there are several drawbacks and difficulties. A transcutaneous passage device or telemetric techniques are required to transfer the sensory information from the outside of the body to the inside. The electric stimulation of sensory fascicles may not be modality-specific and may give rise to unphysiological and weird sensory perception. The nerve interface principle, based on implanted electronic devices will therefore remain on the experimental stage for many years to come.
Attempts to use transferred cutaneous stimulation to remote skin areas were already tried several decades ago. According to this principle remote skin areas of the body can be subjected to electro-cutaneous stimulation (Szeto and Riso, 1990) or vibration (Mann and Reimers, 1970). Although such types of stimuli, when proportional to the pressure which is applied to sensors in the prostheses, can serve as a feedback system for regulation of the grip force in a myoelectric prostheses (Lundborg et al., 1999) they give rise to very unpleasant and non-physiological sensations. In addition, stimulation of remote skin areas is not easy to correlate with stimulation of the hand since the cortical representations of the respective body parts are widely apart and can therefore not serve as a useful system for conscious sensory feedback. No attempts have been made to correlate stimulation of topographically specific skin areas to stimuli applied to analogous topographically defined components of a hand prosthesis.
Amputation of a hand has immediate and long-lasting effects on the functional organisation of brain cortex (Merzenich and Jenkins, 1993; Weiss et al., 2000). Normally the cortical “body map” constitutes representational areas of all body parts which are represented in somatosensory cortex in a specific topographical order and in sizes which are proportional to their sensory competence (Kaas, 1997). Thus, the hand and the face, presenting exceptionally well developed sensory functions, occupy a major part of the somatosensory cortex. According to the cortical body map the face representational area is localised immediately caudal to the hand representation, and the representational area of the forearm is localised immediately cranial to the hand representational area (Penfield and Boldrey, 1937; Merzenich et al., 1987; Kaas, 1997). If a hand is amputated there suddenly is a “silent” cortical area, previously devoted to the hand. This is followed by a rapid expansion of the face and forearm representational areas to include the former hand representational area. It has been shown that this phenomenon may already occur within 24 hours after amputation (Borsook et al., 1998). Touch of the forearm or the face then results in sensory phenomena in the phantom limb as a result of cortical expansion of the forearm representation. Thus, a “phantom hand” can sometimes usually be mapped in the distal part of the residual forearm of the amputee (Ramachandran, 1998). Such cortical functional reorganisation phenomena—an expression for “brain plasticity”—occur rapidly and may remain permanently.
The same problem with lack of conscious sensory feedback also applies to body parts lacking sensation or having impaired sensation, e.g. diabetes patients.